Comparative Analysis of Traditional Machine Learning Approaches for Time Series Clustering Under Colored Noise

The work examines time series clustering using various machine learning approaches. The purpose of the study is to compare the performance of K-shape, K-means, and hierarchical density-based spatial clustering with noise (HDBSCAN) algorithms. Clusters were measured using the Rand Index, Adjusted Rand Index, Adjusted Mutual Information (AMI), and measures based on the electrocardiogram (ECG), ArrowHead, and SharePriceIncrease datasets. Noise was added to the time series data and clustering was performed on the noisy data. The resulting clusters were compared to the original clusters using Bcubed metrics to evaluate the robustness and accuracy of clustering algorithms in noise engineering. The results of the study will shed light on the effectiveness of these clustering algorithms in detecting anomalies in time series data. Additionally, the influence of colored noise on the accuracy and stability of the clustering algorithm will be determined.